基于支持向量回归机的汽轮机排汽焓预测研究  

Study of the Prediction of the Exhaust Steam Enthalpy of a Steam Turbine Based on the Supporting Vector Regression Machine

作　　者：米兰 王文斌 

机构地区：[1]乌海职业技术学院,内蒙古乌海016000 [2]雅砻江流域水电开发有限公司,四川成都610051

出　　处：《热能动力工程》2016年第11期38-42,共5页Journal of Engineering for Thermal Energy and Power

基　　金："十二五"国家科技支撑计划项目(2014BAD06B00)

摘　　要：针对现有汽轮机排汽焓计算方法尤其是神经网络预测方法存在的网络结构不好确定、易陷入局部极值等不足,本文提出了一种新的基于支持向量回归机的汽轮机排汽焓预测方法。在分析了汽轮机排汽焓预测主要影响因素的基础上,为了简化计算流程提高预测的有效性,输入参数中排除了也可能处于湿蒸汽区的7、8段抽汽焓,进一步构建了基于支持向量回归机的汽轮机排汽焓预测模型。应用实例的仿真结果表明,该方法具有较强的泛化能力,可以快速、准确地完成汽轮机排汽焓在线预测。In the light of such demerits existing in the prevailing steam turbine exhaust steam enthalpy calculation method,especially the neural network prediction method,as difficult to determine the network structure and easy to meet with a local extremum etc.,proposed was a new steam turbine exhaust steam enthalpy prediction method based on the supporting vector regression machine. On the basis of analyzing the main factors influencing the prediction of the exhaust steam enthalpy of a steam turbine,to simplify the calculation flow path and enhance the prediction efficiency,the extraction steam enthalpy in the section No. 7 and 8 was excluded in the input parameters,which possibly located in the wet steam zone. On this basis,a model for predicting the exhaust steam enthalpy of a steam turbine was established based on the supporting vector regression machine. The simulation results in the cases show that the method in question has a relatively strong generalization ability and can quickly and accurately fulfill an on-line prediction of the exhaust steam enthalpy of a steam turbine.

关 键 词：汽轮机 排汽焓 支持向量机 支持向量回归机 

分 类 号：TK262[动力工程及工程热物理—动力机械及工程]